# Fix list: Axolotl From Anchor Terminal's listing at https://www.anchorterminal.com/tools/axolotl, the October 2026 research run, assessed 8 October 2026. Grade B, 64.8 out of 100. This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public. For a coding agent working on Axolotl: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published. ## 1. Security & auth, 52 out of 100, up to 8.4 more on the total Why it scored 52: Read as local software. No account or key of its own; Hugging Face and tracker tokens come from the environment (20 of 30). No read-only or approval mode, `trust_remote_code` is off unless the config sets it, and a config path may be a remote URL (8 of 20). A trainer returns no untrusted content to the agent (10 of 15). Runs log locally and to Weights & Biases, MLflow or Trackio when configured, with no audit log of its own (8 of 15). `.github/SECURITY.md` gives an email address for reports and supports only the latest release; no security.txt, bounty, certification or published advisory found (6 of 20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-security): - 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option. - 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions. - 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10. - 0 to 15, audit logs or per-call visibility for the operator. - 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public. Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing. ## 2. Reliability, 64 out of 100, up to 7.2 more on the total Why it scored 64: Scored on the local-software lines. Installs from PyPI (`axolotl` 0.20.0) or the `axolotlai/axolotl` Docker image, with Python 3.12 or later and PyTorch 2.13 or later stated (20). Public GitHub Actions and 536 test files; of the last six push runs of the Tests workflow on main, three passed, two failed and one was cancelled, and the latest on 8 October failed (15 of 25). 109 open issues against 199 commits since 10 July, two of the 15 newest labelled bug (17 of 25). Release notes carry a Deprecations section naming removals, but 0.x minors break configs, such as `relora_steps` renamed in 0.17.0 with no shim (12 of 15). Version 0.20.0, no stability declaration (0). The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability): Hosted APIs, MCP servers, models and platforms. - 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own). - 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so. - 15, rate limits documented with numbers. - 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved. - 10, an SLA published for any paid tier. - 10, the surface agents use is generally available, not beta or preview. Local packages, SDKs, frameworks and stdio MCP servers. - 20, installs from an official package with supported runtimes stated. - 25, a public CI and test suite, passing on the default branch. - 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered). - 15, semver discipline and breaking changes called out in a changelog. - 15, version 1.0 or later, or declared stable. Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors. ## 3. Agent ergonomics, 60 out of 100, up to 6.5 more on the total Why it scored 60: Read as a CLI. One YAML file drives preprocess, train, evaluate, merge, quantise and export, and agent docs load one topic at a time (18 of 25). Output is sized by saving an adapter or a merged model and by choosing export quantisation, with `--debug-num-examples` for previews (10 of 20). Config validation raises named errors and there is a debugging guide; error text was not tested in this run (12 of 20). `--resume-from-checkpoint` is the safe retry (12 of 20). Example configs and CLI overrides for common fields, Python only (8 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics): - 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries). - 20, pagination, filtering and output-size controls. - 20, actionable, documented error responses, codes and messages an agent can recover from. - 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations. - 15, sensible defaults, few required parameters, and official SDKs in at least two languages. Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs. ## 4. Payments & pricing, 60 out of 100, up to 5 more on the total Why it scored 60: Scored by the self-hosted rule. Free software with no price list or hosted plan on axolotl.ai, so 20 + 20 + 20 for public pricing, free use and no sign-up. Dedicated support is by email with no published price. No x402, MPP or L402 (0). The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments): The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/). - 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which. - 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login. - 20, a free tier or trial that doesn't need a card. - 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API). Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied. Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol. ## 5. Transparency & trust, 56 out of 100, up to 3.9 more on the total Made of editorial 75, provenance 36. Why it scored 56: Apache-2.0 in the repository, though the PyPI metadata carries no licence field (30). Training data stays on the owner's machine and the telemetry page lists what is sent to PostHog, but there is no privacy policy and no retention period (15 of 30). Deprecations are named in release notes and marked in the support matrix, with no dated removal policy (10 of 20). Telemetry is on by default, disclosed in the README and docs, announced at start-up, and turned off with `AXOLOTL_DO_NOT_TRACK=1` or `DO_NOT_TRACK=1` (20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency): - 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms. - 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors). - 0 to 20, a deprecation policy or notices with dates. - 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted). The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two. Provenance checks not met in full (half of this category, computed from checked facts): - Legal entity named: not found (0 of 20) - Domain age: axolotl.ai, registered 2022-08-02 (4 years) (7 of 15) - Status page: not found (0 of 10) - security.txt: not found (0 of 10) ## 6. Schema & documentation, 80 out of 100, up to 3.3 more on the total Why it scored 80: Read as a framework an agent drives through a CLI. `axolotl config-schema` prints the Pydantic config as JSON Schema, whole or per field, and the docs carry a config reference and a generated API reference (20 of 25). No llms.txt on either host (404), but `axolotl agent-docs` prints Markdown references bundled in the package (8 of 10). A method-choice guide and a support matrix mark each option stable, experimental or deprecated (15 of 20). Typed config fields with defaults and validators (12 of 15). An examples folder by model family, a debugging guide and an FAQ (12 of 15). Tagged releases with long notes, no changelog file (13 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema): APIs and MCP servers. - 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool). - 10, llms.txt or Markdown docs served for agents. - 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference. - 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs. - 0 to 15, examples and documented error responses. - 15, versioning and a public changelog. Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference. ## 7. Maintenance & community, 88 out of 100, up to 1.1 more on the total Why it scored 88: Version 0.20.0 on PyPI on 30 September 2026 (30). Three releases since 10 July, 0.18.0 on 17 July, 0.19.0 on 10 September and 0.20.0 on 30 September (20). 199 commits since 10 July and a commit on 8 October; 109 open issues, and nine of the 15 newest had no reply (16 of 25). Current PyPI package and Docker images (15). Dependencies pinned and updated by Dependabot, with mixed CI results on main this week (7 of 10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance): - 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older. - 20, at least three releases or dated changelog entries in the last 90 days. - 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15. - 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models). - 10, package health, current dependencies and CI. Models are read for deprecation notice periods and model churn rather than release counts. ## What we couldn't check What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it. - The legal entity behind Axolotl AI was not found on axolotl.ai, in the docs or in the repository. - No terms of service or privacy policy is published (axolotl.ai/terms and /privacy return 404), so `provenance.terms` and `provenance.privacy` are left out. - unchecked: the output of `axolotl config-schema` and the CLI's error messages were read in source, not run. - unchecked: the telemetry retention period and who can read the PostHog project. - unchecked: why the Tests workflow failed on main on 8 October 2026; the run logs were not read. - The repository's `AGENTS.md` and `CLAUDE.md` carry instructions addressed to AI coding agents about contributions. Recorded as a fact; none was acted on. ## Weaknesses - Telemetry to PostHog is on by default and delays training start by 10 seconds until the variable is set either way - No terms of service, privacy policy, legal entity or security.txt found on axolotl.ai - Version 0.20.0, with removals in minor releases (FSDP1 in 0.20.0, `relora_steps` renamed in 0.17.0 with no shim) - Three of the last six push runs of the Tests workflow on main passed, and the nightly run against upstream failed on 7 and 8 October 2026 - Not a hosted service, so there is no job API, status page or SLA ## What costs an agent a turn today The notes we give agents before they call it. Each one is a workaround an agent shouldn't need. - Set `AXOLOTL_DO_NOT_TRACK=1` before any command, or training waits 10 seconds and sends usage events to PostHog - Run `axolotl agent-docs` and `axolotl config-schema --field ` before writing a config; both work offline from the installed package - Install torch first, then `uv pip install --no-build-isolation axolotl[deepspeed]`, on Python 3.12 or later with PyTorch 2.13 or later - Take example configs from the same release tag as the installed version; minor releases remove and rename config keys - Resume an interrupted run with `axolotl train config.yml --resume-from-checkpoint `, then `axolotl merge-lora` and `axolotl export` only when shipping ## When it's done Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.